arXiv:astro-ph/0311562v2 12 Jan 2004 nohrmdl o h D eegvni e.[21]. Ref. in given were CDM the estimates Ref. for Analytic See models description. scenario. other theory this in in field matter a the in by equations inherited be may that easit o-eaiitccl akmte CM.Ti a This three-fl (CDM). a matter to dark analysis cold this m non-relativistic extend two-fluid into we a decays paper for b this [22] In has in what performed radiation. was in This time equations. decay bation the at res density the curvaton of the calculation approximate flui of an terms the derive in as individual to long used define so been can conserved has One approximately remain 21]. that 20, fluids 15, isoc [14, residual of scenario possibility curvaton the open leaves perturbations 1,19]. [18, n h oa uvtr perturbation curvature total the and rirrl ag cls[0.Atog ti fe assumed often infl the is driving it field sma scalar Although slowly-rolling where a era of [10]. fluctuations inflationary scales earlier large an perturb arbitrarily in primordial these fluctuations that quantum is Gaussi explanation adiabatic, attractive of Universe An spectrum the primordial of a picture with our consistent improved greatly have periments etraindpnsuo hte rnttelclvleo t of value local the 12]. not fluctuati [11, or density isocurvature whether alm these upon an not depends acquire or perturbation will Whether scale) Hubble expansion. the tial than less mass effective osrit:oecncluaetepioda perturbatio primordial the calculate can one constraints: upon t constraints leads different isocu qualitatively scenario initial and This the [16] is inflation perturbation. it and density den inflation primordial radiation of the local end the the at Then curvaton co universe. the to the decays. On enough of eventually density large and grown energy inflation. have oscillates of may field curvaton end the non-relativistic the curvaton, after the decoupled of remains and inflation omlg nlresae n n a en osre curva conserved a define can one and scales large on cosmology r.Ti sbcueteifltnflcutosdescribe fluctuations inflaton the because i of is end This the of physics era. detailed the of independently flation, nscinI estu h ope vlto qain o s a for equations evolution coupled the up set we II section In oee nacrt aclto ftersligprimordia resulting the of calculation accurate an However ycnrs,flcutosi h uvtnfil describe field curvaton the in fluctuations contrast, By bevtoa eut bandb ait frcn cosmi recent of variety a by obtained results Observational ntecrao cnro[3 4 5 ih,wal coupled weakly light, a 15] 14, [13, scenario curvaton the In ntecnetoa lwrl nao cnroteei re a is there scenario inflaton slow-roll conventional the In uvtr n scraueprubtosi he-udm three-fluid a in perturbations isocurvature and Curvature 1 nttt fCsooyadGaiain nvriyo Port of University Gravitation, and Cosmology of Institute 2 RC,Isiu ’srpyiu ePrs .... 98bis C.N.R.S., Paris, de d’Astrophysique Institut GRECO, n oddr atratrtecrao a eae.W iet 98.80.Cq give decay We numbers: curvaton PACS the decayed. to has relative curvaton density the curvaton after initial matter the dark isocurvatu cold correlated and and curvature a primordial radiation the both culate into decays curvaton non-relativistic the esuyteeouino h omlgclprubtosaf perturbations cosmological the of evolution the study We 3 hsc eatet acse nvriy acse A 4 LA1 Lancaster University, Lancaster Department, Physics uaaGupta Sujata ζ sn ogrcnevdo ag cls h o-daai na non-adiabatic The scales. large on conserved longer no is 1 ai .Malik A. Karim , Dtd eray2 2008) 2, February (Dated: .INTRODUCTION I. raueprubtosaddtcal o-asint in non-Gaussianity detectable and perturbations urvature adiabatic sdrcl ntrso h nao edflcutosdrn i during fluctuations field inflaton the of terms in directly ns non-adiabatic toayepnin(h inflaton), (the expansion ationary h nainr aaees[17]. parameters inflationary the n cl-needn etrain 1 ,3 ,5 ,7 ,9 8, 7, 6, 5, 4, 3, 2, [1, perturbations scale-independent an, flto,rhaigadtesbeun radiation-dominate subsequent the and reheating nflation, tosi h al aito-oiae r rgnt from originate era radiation-dominated early the in ations fteapiueo h eiuliouvtr perturbation isocurvature residual the of amplitude the of i ytmt nld h osblt httecrao also curvaton the that possibility the include to system uid lw st aclt h eiuliouvtr perturbati isocurvature residual the calculate to us llows e osblte o uligpril hsc oesof models physics particle building for possibilities new o vtr etraino h uvtnfil htdetermines that field curvaton the of perturbation rvature n nnnifltnfilspoueapioda curvature primordial a produce fields non-inflaton in ons iyatrcrao ea eed pntelclvleof value local the upon depends decay curvaton after sity etrain eursnmrclslto ftepertur- the of solution numerical requires perturbations l u ftedcyi aeeog hntedniyo the of density the then enough late is decay the if But tiuesgicnl mr hnaot1)t h total the to 1%) about than (more significantly ntribute htpioda etrain rgnt rmadiabatic from originate perturbations primordial that ligpioda etrain ntecrao scenario curvaton the in perturbations primordial ulting ntelretosral cls hs bevtosare observations These scales. observable largest the on s cl-nain pcrmdrn nams exponen- almost an during spectrum scale-invariant ost dlweennrltvsi uvtnprilsdcyinto decay particles curvaton non-relativistic where odel uvtr etrain o h aito n curvaton and radiation the for perturbations curvature akbesmlfiainwe acltn observational calculating when simplification markable scnb sue ob o-neatn 1,1] This 19]. [18, non-interacting be to assumed be can ds clrfil ean eope rmteifltnduring inflaton the from decoupled remains field scalar irwv akrudadlresaesrcueex- structure scale large and background microwave c a clrfil usqetyaet h oa radiation local the affects subsequently field scalar hat 1]franmrclslto fteprubdcurvaton perturbed the of solution numerical a for [15] ueperturbation ture 2 muh otmuhP12G ntdKingdom United 2EG, PO1 Portsmouth smouth, lsaevcu utain a esrthdu to up stretched be can fluctuations vacuum scale ll , eteHbl cl rp eo h ffciemass effective the below drops scale Hubble the ce 3 e aldtesde-ea prxmto 1,20]. [14, approximation sudden-decay the called een dacl akmte opnn.W cal- We component. matter dark cold a nd eprubtosihrtdb h radiation the by inherited perturbations re etrain bu h akrudhomogeneous background the about perturbations rate. n ai Wands David and se fdmninesvralsdsrbn the describing variables dimensionless of ystem e naini uvtnmdl where models curvaton in inflation ter olvr rg,704Prs France Paris, 75014 Arago, Boulevard etrain bu h akrudcosmology background the about perturbations UIG0/3 sr-h0152v2 astro-ph/0311562 PU-ICG-03/43, etase offiin ntrsof terms in coefficient transfer he ζ B UK YB, ietyfo nryconservation energy from directly dlo uvtndecay curvaton of odel 1 any ih clrfil (with field scalar light ueo h initial the of ture the on n- ]. d 2 evolution of the background homogeneous cosmology. We then give the coupled evolution equations for linear density perturbations in each fluid on uniform-curvature hypersurfaces in section III. In an appendix we present equations describing the evolution of curvature and isocurvature perturbations in manifestly gauge-invariant form and discuss the appearance of singularities for some gauge-invariant quantities in a system such as this that includes energy transfer between fluids. This is seen to be due to a singular choice of hypersurface rather than a breakdown of perturbation theory. The primordial perturbations that result from an initial curvaton perturbation can be characterised by a transfer coefficient that is a unique function of a single dimensionless parameter that specifies the initial density of the curvaton relative to the curvaton decay rate. We give an accurate analytic approximation (to better than 1%) for the curvature transfer coefficient. The same coefficient also determines non-Gaussianity of the primordial perturbations [14] and, in our three-fluid model, the amplitude of the residual isocurvature perturbation. We are able to derive, as an exact result, a simple consistency relation between the ratio of isocurvature to curvature perturbations and the non-Gaussianity of the primordial perturbations, which was previously derived only using the sudden-decay approximation. We present our conclusions in section IV.
II. BACKGROUND EQUATIONS AND DECAY
In this section we will first review the governing equations of the background quantities in a Friedman–Robertson– Walker (FRW) universe and then find the evolution equations for the perturbations for three interacting fluids, following Ref. [22]. We begin the analysis after the end of inflation when the vacuum energy driving inflation has decayed into radiation. We assume that at that time the curvaton density is much smaller than the total density, consistent with the assumption that curvaton fluctuations are initially isocurvature perturbations. The radiation (γ) is a relativistic fluid with pressure Pγ = ργ /3. We model the other two components, the curvaton (σ) and the cold dark matter (m), as non-relativistic fluids, Pσ = Pm = 0. This is a good description of the classical field dynamics once the Hubble rate has dropped below the effective mass of the curvaton so that the curvaton field begins coherent oscillations. The evolution of a flat background FRW universe is governed by the Friedmann constraint and the continuity equation, respectively, 8πG H2 = ρ , (1) 3 ρ˙ = −3H (ρ + P ) , (2) where the dot denotes a derivative with respect to coordinate time t, H is the Hubble parameter, and ρ and P are the total energy density and the total pressure, related to the density and pressure of the component fluids by
ρα = ρ , Pα = P . (3) α α X X The continuity equation for each individual fluid in the background can be written as [23]
ρ˙α = −3H (ρα + Pα)+ Qα , (4) where Qα describes the energy transfer per unit time to the α-fluid. We assume the curvaton can decay into both radiation and cold dark matter, described by the two decay rates, Γ1 and Γ2 respectively. For simplicity we assume the decay rates are fixed constants. This gives the background energy transfer as
Qσ = −(Γ1 +Γ2)ρσ , (5)
Qγ = Γ1ρσ , (6)
Qm = Γ2ρσ . (7) The cold dark matter in this model is assumed to be a non-relativistic product of curvaton decay, which does not interact with radiation even at these early times. Γ2/Γ1 determines the energy density of cold dark matter particles relative to the radiation produced by curvaton decay. For the CDM to remain subdominant until matter-radiation equality at Teq ≃ 1eV we require
Γ Ω T − 2 . m = eq < 10 6 . (8) Γ Ω T 1 γ decay decay 3
The final inequality follows from requiring that the curvaton decays before primordial nucleosynthesis (Tdecay > 1MeV), after which time any significant entropy production would spoil theoretical predictions of the primordial abundance of light elements. The background energy transfer must always obey the constraint
Qα =0 . (9) α X It is useful to rewrite the background equations (4), in terms of dimensionless density parameters ρ ρ ρ Ω ≡ σ , Ω ≡ γ , Ω ≡ m , (10) σ ρ γ ρ m ρ which gives the evolution equations
′ Γ +Γ Ω = Ω Ω − 1 2 , (11) σ σ γ H ′ Γ Ω = Ω 1 +Ω (Ω − 1) , (12) γ σ H γ γ ′ Γ Ω = Ω 2 +Ω Ω , (13) m σ H m γ where a dash denotes the differentiation with respect to the number of e-foldings, N ≡ Hdt. The Friedmann constraint (1) then yields R Ωσ +Ωγ +Ωm =1. (14) The evolution of the inverse Hubble parameter is governed by ′ 1 1 3 = 1+ Ω . (15) H 3 γ 2H Hence we have an autonomous dynamical system consisting of the four Eqs. (11–13), and (15), but the constraint (14) makes one of them redundant.
A. Numerical solutions
In Fig.1 we show a phase-plane which plots the evolution of the curvaton energy density against the dimensionless decay rate, 0 < Γ1/(Γ1 + H) < 1. Note that for Γ2 ≪ Γ1 the evolution of Ωσ is effectively independent of Γ2 and the phase-plane shown is identical to that studied in Ref.[22]. All trajectories in the phase-plane in Fig.1 start close to the origin with Ωσ ≪ 1 and H ≫ Γ1. From the limiting −1/2 N form of Eqs. (11) and (14) for Ωγ ≃ 1 we can see that Ωσ ∝ H ∝ e [22]. Thus different trajectories can be identified by different values of the dimensionless parameter
1 H 2 p ≡ Ωσ . (16) Γ1 " #initial
If we consider the initial conditions for the fluid model to be set when the curvaton begins to oscillate (Hin = mσ) then we can write
8πhσ2i m 1/2 p = in σ . (17) 3M 2 Γ Pl 1 2 2 The curvaton vacuum expectation value is bounded, hσ iin 1 A 0.8 0.6 B σ Ω 0.4 0.2 C 0 0 0.2 0.4 0.6 0.8 1 Γ1 /(Γ 1+ Η ) FIG. 1: The phase plot shows the evolution of the dimensionless energy density, Ωσ , for different values of the dimensionless 3 −3 parameter, p, defined in Eq. (16), varying from 3.15 × 10 (top line) to 3.15 × 10 (bottom line). For all of the lines Γ2 ≪ Γ1. The evolution of the densities of all three fluids are plotted below in Figs. 2 to 4 for the trajectories, marked A, B and C. energy density of the curvaton then rapidly decays when H ∼ Γ1. In this example there is a period of curvaton domination (Ωσ ≃ 1) before the curvaton decays. Figure 3(a), corresponding to trajectory B of the phase–plane figure (1), has a lower value the parameter p, so the curvaton density does not climb as high before it decays away. Figure 4(a), corresponding to trajectory C of the phase–plane figure (1), has the lowest value of p of the three plots and the ratio of initial curvaton density to the decay rate is low enough that there is no period of curvaton domination in this example. III. PERTURBED EQUATIONS The evolution equation for density perturbations in the α fluid is given on large scales and on uniform-curvature hypersurfaces by [22] δρ δρ˙ +3H(δρ + δP )+ Q − δQ =0 , (18) α α α α 2ρ α where δρα, δPα and δQα are the density perturbations, the pressure perturbation, and the energy transfer perturbation of the α-fluid. δρ is the total density perturbation on the uniform-curvature hypersurfaces and appears in the equation for the perturbed energy transfer due to the gravitational time dilation [22]. The perturbed energy transfer (assuming the physical decay rates are unperturbed) is given by, δQσ = −(Γ1 +Γ2)δρσ , (19) δQγ = Γ1δρσ , (20) δQm = Γ2δρσ . (21) We will evolve the perturbed conservation equations (18), which can be written in terms of the dimensionless density 5 1 0.8 Ω 0.6 σ Ω γ Ω 0.4 m 0.2 0 0 5 10 15 20 25 N (a) 5 ζ ζ / σ,in 4 ζ ζ m/ σ,in ζ ζ γ/ σ,in ζ ζ 3 σ/ σ,in 2 1 0 0 5 10 15 20 25 N (b) FIG. 2: (a) The evolution of Ωσ (solid), Ωγ (dashed) and Ωm (dotted) against e-foldings N of line A of Fig.1. The initial −2.5 −10 −12 value of Ωσ is 10 , and Γ1 = 10 and Γ2 = 10 . (b) The evolution of ζσ/ζσ,in (dotted), ζγ /ζσ,in (dot–dashed), ζm/ζσ,in (dashed) and the total curvature perturbation, ζ/ζσ,in (solid), against e-foldings N of line A of Fig.1. parameters as ′ Γ +Γ Γ +Γ δρ δρ + 3+ 1 2 δρ − 1 2 Ω =0 , (22) σ H σ H σ 2 ′ Γ δρ δρ +4δρ − 1 δρ − Ω =0 , (23) γ γ H σ σ 2 ′ Γ δρ δρ +3δρ − 2 δρ − Ω =0 . (24) m m H σ σ 2 where a prime denotes derivatives with respect to N. Written in this form it is evident that the linear evolution of the density perturbations depends only upon the dimensionless background variables Ωα and decay rates Γα/H. Hence the final amplitude of the density perturbations, δρα,final, relative to the initial curvaton perturbation δρσ,initial will depend only upon which trajectory is followed in the background phase-space (Ωα, Γα/H). They will not depend on dimensional quantities such as Hinitial. In fact we will show that the large-scale perturbations after the curvaton has decayed can be written simply in terms of the initial curvaton perturbation and a function of the dimensionless parameter p defined in Eq. (16). 6 1 0.8 Ω 0.6 σ Ω γ Ω 0.4 m 0.2 0 0 5 10 15 20 25 N (a) 5 ζ ζ / σ,in 4 ζ ζ m/ σ,in ζ ζ γ/ σ,in ζ ζ 3 σ/ σ,in 2 1 0 0 5 10 15 20 25 N (b) FIG. 3: (a) The evolution of Ωσ (solid), Ωγ (dashed) and Ωm (dotted) against e-foldings N of line B of Fig.1. The initial value −4.6 of Ωσ is 10 . Γ1 and Γ2 are the same as in Fig.2. (b) The evolution of ζσ/ζσ,in (dotted), ζγ /ζσ,in (dot–dashed), ζm/ζσ,in (dashed) and the total curvature perturbation, ζ/ζσ,in (solid), against e-foldings N of line B of Fig.1. A. Curvature and isocurvature perturbations The gauge-invariant variable, ζ, introduced by Bardeen [24, 25], is commonly used to describe the large-scale curvature perturbation. It is equivalent to the curvature perturbation of uniform-density hypersurfaces and remains constant on super–horizon scales for adiabatic perturbations. This quantity is simply related to the total density perturbation on uniform-curvature hypersurfaces [18, 25] δρ ζ = −H . (25) ρ˙ We can also define the curvature perturbation on uniform α-fluid density hypersurfaces, ζα, which is related to the α-fluid density on uniform-curvature hypersurfaces [18] δρα ζα = −H . (26) ρ˙α The total density perturbation is just the sum of the density perturbations of the individual fluids, δρ = α δρα, and hence the total curvature perturbation ζ is the weighted sum of the individual perturbations [18], P ρ˙ ζ = α ζ . (27) ρ˙ α α X 7 1 0.8 Ω 0.6 σ Ω γ Ω 0.4 m 0.2 0 0 5 10 15 20 25 N (a) 5 ζ ζ / σ,in 4 ζ ζ m/ σ,in ζ ζ γ/ σ,in ζ ζ 3 σ/ σ,in 2 1 0 0 5 10 15 20 25 N (b) FIG. 4: (a) The evolution of Ωσ (solid), Ωγ (dashed) and Ωm (dotted) against e-foldings N of line C of Fig.1. The initial value −5.5 of Ωσ is 10 . Γ1 and Γ2 are the same as in Fig.2. (b) The evolution of ζσ/ζσ,in (dotted), ζγ /ζσ,in (dot–dashed), ζm/ζσ,in (dashed) and the total curvature perturbation, ζ/ζσ,in (solid), against e-foldings N of line C of Fig.1. The difference between any two individual curvature perturbations, ζα and ζβ, describes an isocurvature or entropy perturbation [18, 22] Sαβ = 3 (ζα − ζβ) . (28) Note that the definition of ζα becomes singular wheneverρ ˙α = 0. For instance the physical density of cold dark matter is initially vanishing, but growing due to the decay of the curvaton (ρ ˙m > 0). However after the curvaton has completely decayed, the matter density is diluted by the expansion (ρ ˙m < 0). At one instant we haveρ ˙m = 0 and ζm is ill-defined. This has been previously noted in the case of an oscillating scalar field at stationary values of the field [26]. It is a consequence of the uniform α-density hypersurfaces becoming ill-defined, and does not signal the breakdown of perturbation theory, as long as we can find some gauge where the hypersurfaces are well-defined and perturbations remain small (see Appendix). In this case we will numerically solve for the fluid density perturbations on uniform-curvature hypersurfaces. Fig. 5 shows an example of the evolution of the density perturbations in equations (22) to (24) above, showing that they do indeed remain well-defined throughout the evolution. From the density perturbations we then construct the curvature perturbations, Eq.(25) and (26), which gives 8πG δρσ ζσ = , (29) 3H Ωσ (3H +Γ1 +Γ2) 8 δρ 1e-10 σ δρ γ δρ 1e-15 m 1e-20 1e-25 1e-30 1e-35 0 5 10 15 20 25 N FIG. 5: The evolution of the density perturbations on uniform-curvature hypersurfaces, δρσ (solid), δργ (dashed) and δρm (dotted), against e-foldings N along the background trajectory B in Fig.1. 8πG δργ ζγ = − , (30) 3H (−4HΩγ +Γ1Ωσ) 8πG δρm ζm = − , (31) 3H (−3HΩm +Γ2Ωσ) At late times, after the curvaton has completely decayed the fluids are non-interacting and have monotonically decreasing energy densities (diluted by the expansion) so the curvature perturbations, ζα, are well-defined, and constant on large scales. Observations of, for example, the CMB directly constrain the amplitude of the total curvature and isocurvature perturbations. On large angular scales the temperature perturbations on the surface of last-scattering are given by the Sachs-Wolfe effect [6, 7, 27] δT 1 = (−ζ − 2S ) . (32) T 5 γ mγ lss lss B. Numerical results We have numerically evolved the coupled system of background equations (11)–(13) and (15) and first-order per- turbations (22)–(24). In each case we begin with a smooth (ζγ,in = 0), dominant radiation fluid (Ωγ,in ≃ 1) and a subdominant curvaton fluid (Ωσ,in ≪ 1) with an initial perturbation ζσ,in 6= 0. The cold dark matter density is initially zero (Ωm,in = 0, ζm,in = 0), and is produced solely from curvaton decay. The initial density perturbations on uniform-curvature hypersurfaces are given in terms of ζα,in from Eqs. (29)–(31). In all cases we start the evolution at some high energy scale, with a large value of the Hubble parameter, characterised by a small initial value of the Hubble time relative to the decay time, Γ1/Hin ≪ 1. The subsequent evolution is not directly affected by the precise value of Γ1/Hin, but rather by the parameter p defined in Eq. (16) which characterises the phase-space trajectory. In particular the final values of the curvature and isocurvature perturbations are a function solely of the parameter p. Figures 2(b), 3(b) and 4(b) show the evolution of the large–scale curvature perturbations for the three fluids along three different background trajectories. 1. Radiation The radiation curvature perturbation ζγ grows as the curvaton decays, and reaches an equilibrium point in the phase-space where ζγ = ζσ [22]. The final value for ζγ relative to the initial value of ζσ,in depends upon the relative density of the curvaton at the decay epoch, Ωσ,decay. This determines the proportion of the final radiation that comes from the decay of the perturbed curvaton. 9 1 r 0.1 0.01 0.1 1 10 100 p 1/2 FIG. 6: Curvature transfer coefficient r ≡ ζγ,out/ζσ,in, plotted against background parameter p ≡ Ωσ (H/Γ1) . in h i For example, in Fig. 2(a), corresponding to trajectory A of Fig. 1, the curvaton comes to dominate (Ωσ ≃ 1) before the decay epoch. Thus almost all the radiation at late times comes from the decay of the curvaton and hence, in Fig. 2(b), the radiation fluid at late times inherits the initial curvaton perturbation, ζγ ≃ ζσ,in. Compare this with Fig. 4, corresponding to trajectory C of the Fig. 1. In this case the curvaton decays while Ωσ ≪ 1 and the proportion of the final radiation energy density which results from the decay of the perturbed curvaton component is much smaller in this picture. We define r to be the ratio between the curvature perturbation of the radiation component at the end of the calculation (when all of the curvaton component has decayed away) to the initial curvaton perturbation, ζγ,out = rζσ,in . (33) We find that r is solely a function of the choice of trajectory in the background phase-space, Fig. (1). Each trajectory is uniquely identified in terms of the initial data by the value of the parameter p defined in Eq. (16). We plot r against p in Fig. (6). At small values of p this approaches a linear function r ≃ 0.924p , (34) see Fig. (7). This coincides with the sudden-decay approximation calculated in Ref. [22] for Ωσ,in ≪ 1. The linear approximation is good to better than 1% up to p =0.011 (r =0.0108) and better than 10% up to p =0.11 (r =0.10). At larger values of p the transfer coefficient r(p) turns over and r → 1 for p ≫ 1. The full function is well fit by − 0.924 q r (p)=1 − 1+ p , (35) fit q where, by fitting over the range 10−2 2. Cold Dark Matter In our model the universe starts with an initial cold dark matter density, ρm, equal to zero which then grows due to curvaton decay, but eventually starts decreasing due to cosmic expansion. As remarked earlier the curvature perturbation associated with the cold dark matter, ζm, becomes ill-defined whenρ ˙m = 0. This occurs when Qm = Γ2ρσ =3Hρm, in Eq. (4) for α = m. The divergence of ζm is clearly seen at early times (before the curvaton decays) in Figures 2(b), 3(b) and 4(b). The curvature perturbation ζm is initially time-dependent due to the energy transfer from the perturbed curvaton, but settles down to a constant value at late times, once the curvaton has completely decayed Γ1 +Γ2 ≫ H. We find that if the cold dark matter is produced solely from curvaton decay, it inherits a curvature perturbation ζm,out at late times equal to the initial curvaton perturbation, ζσ,in. 10 1 0.1 r 0.01 0.01 0.1 1 p FIG. 7: Curvature transfer coefficient r, defined in Eq. (33), plotted against parameter p, defined in Eq. (16), with a linear fit r = 0.924p for p ≪ 1. This linear fit is accurate to within 1% for p< 0.011. 1 1e-1 1e-2 1-r 1e-3 1e-4 1 10 100 1000 p FIG. 8: Isocurvature transfer coefficient, 1−r, defined in Eq. (33), plotted against parameter p, defined in Eq. (16). Also shown (dot-dashed line) is the fitting function 1 − rfit, defined in Eq. (35). This is accurate to within 1% for p< 8.14 (corresponding to 1 − r > 0.086). This simple result can be derived by considering a composite energy density constructed from the CDM density and a fixed fraction of the curvaton density: Γ2 ρcomp ≡ ρm + ρσ . (36) Γ1 +Γ2 From Eqs. (5) and (7) we see that the corresponding energy transfer is zero, i.e., Γ2 Qcomp ≡ Qm + Qσ =0 . (37) Γ1 +Γ2 Thus we obtain from Eqs. (4) and (18) the standard evolution equations for the composite energy density and its perturbation: ρ˙comp = −3Hρcomp , (38) ˙ δρcomp = −3Hδρcomp . (39) Note that as both the curvaton and CDM are pressureless then so is the composite energy density, Pcomp = 0. Since we have a conserved energy density with a unique equation of state the corresponding curvature perturbation, defined by Eq. (26), δρcomp ζcomp = , (40) 3ρcomp 11 is conserved on large scales [18, 19]. Initially ρm = 0 and δρm = 0 and hence ζcomp = ζσ,in. At late times ρm ≫ (Γ2/(Γ1 +Γ2)) ρσ, and ζcomp = ζm ,out. Therefore ζm ,out = ζσ,in , (41) as we see in Figs. 2, 3 and 4. 3. Curvature and isocurvature perturbations After the curvaton has decayed, (Ωσ → 0), we are left with one overall curvature perturbation, Eq. (27), 4Ωγ 3Ωm ζ = ζγ + ζm , (42) 4Ωγ + 3Ωm 4Ωγ + 3Ωm and one relative isocurvature (entropy) perturbation between the radiation and cold dark matter, Eq. (28), Smγ = 3 (ζm − ζγ ) . (43) The primordial curvature perturbation is usually taken to be the curvature perturbation at, for instance, the epoch of primordial nucleosynthesis when the Universe is known to be radiation dominated (Ωγ ≫ Ωm, Ωσ = 0), and hence ζ ≃ ζγ . We have seen that the primordial curvature perturbation relative to the initial curvaton perturbation is a function of the trajectory, r(p) in Eq. (33), ζ|nuc ≡ ζγ,out = rζσ,in , (44) whereas the final cold dark matter curvature perturbation is identical to the initial curvaton perturbation, Eq. (41). Hence S|nuc ≡ Smγ,out = 3(1 − r) ζσ,in . (45) If r = 1 then we obtain an adiabatic curvature perturbation after the curvaton has decayed and there is no late-time isocurvature perturbation. Note that the fitting function rfit(p) given in Eq. (35) gives a good approximation, for 1 − r (better than 1% for r< 0.914 (p< 8.14) or better than 10% for r< 0.993 (p< 66)). Therefore the primordial isocurvature and curvature perturbations are 100% correlated with their ratio fixed by r(p): S 3 (1 − r) = . (46) ζ r nuc This expression was derived in Ref. [20] using the sudden-de cay approximation. We have shown that it is in fact an exact expression when the CDM is produced directly from the decay of the curvaton, and r is defined by Eq. (33). IV. CONCLUSIONS In this paper we have studied the coupled evolution of background densities and linear perturbations of three interacting fluid components: radiation, a curvaton field and cold dark matter. We solve numerically the evolu- tion equations for density perturbations on spatially flat hypersurfaces to calculate the spectrum of curvature and isocurvature (or relative entropy) perturbations that result from initial perturbations in the curvaton field. Whereas the resulting radiation curvature perturbation, ζγ,out, is dependent on the initial values of background parameters (such as curvaton density), we find the simple result that the final CDM curvature perturbation, ζm,out, is always exactly equal to the initial curvaton perturbation ζσ,in, in a model where the CDM is directly produced by curvaton decay. We are able to demonstrate this analytically by constructing a conserved quantity, ζcomp, that at early times is equal to the curvaton perturbation and at late times is equal to the CDM curvature perturbation. In effect this solves the evolution of one degree of freedom and allows us to solve numerically only for ζγ as a function of input parameters. 12 The background evolution of the fluids is described by a trajectories in phase-space distinguished by a dimensionless 1/2 parameter p = [Ω(H/Γ1) ]in. The ratio r between ζγ,out and ζσ,in is a function solely of this trajectory parameter p and is well-fit by a simple numerical function given in Eq. (35). This coincides with the “sudden-decay approximation” [14, 22] for p ≪ 1. The resulting radiation and CDM perturbations after curvaton decay can then be used to predict the spectrum of CMB anisotropies in a given model where the CDM is produced directly as a curvaton decay product. CMB constraints are usually quoted in terms of the primordial curvature ζnuc and an isocurvature perturbation Snuc around the epoch of primordial nucleosynthesis, leading to ζ 1 r/3 ζ = . (47) S 0 (1 − r) S nuc in In the curvaton scenario ζin = ζγ,in = 0 and Sin = 3(ζσ,in − ζγ,in)=3ζσ,in. Although the model we have studied describes the production of CDM directly from the curvaton decay, one would expect a similar result to hold for models where the net baryon or lepton number is directly produced from curvaton decay. The matrix equation (47) is a particular case of the general form of the transfer matrix for curvature and isocurvature perturbations presented in Ref. [6, 28] where the contribution of the initial isocurvature perturbation to the resulting primordial curvature and isocurvature perturbation is model dependent. For instance in a curvaton scenario where the CDM is a thermal relic which decouples from radiation sometime after the curvaton has decayed we have instead ζ 1 r/3 ζ = . (48) S 0 0 S nuc in The matrix equations (47) and (48) using the numerically determined function r(p) [or its approximate form rfit(p)] enable one to relate CMB observations to curvaton model parameters such as the curvaton decay rate, Γ, and the curvaton density, ρσ, at some early time, characterised by the value of the Hubble rate H, and the initial spectrum of curvaton perturbations ζσ = (1/3)δρσ/ρσ. For example, in the model where the CDM is directly produced by curvaton decay we confirm [14] that the relative amplitude of primordial isocurvature to curvature perturbations is a fixed ratio 3(1 − r)/r which is a function solely of the background parameter p defined in Eq. (16). This holds independently of the “sudden-decay approximation” previously used. In any curvaton model the transfer coefficient r is independent of scale in the large-scale limit (by definition) so that the scale-dependence of the primordial perturbation spectra are inherited directly from the scale dependence of the curvaton perturbation at the end of inflation [14, 28] d ln P ∆n ≡ δσ = −2ǫ +2η . (49) σ d ln k σ ˙ 2 2 2 In this expression ǫ ≡ −H/H is the usual dimensionless slow-roll parameter during inflation and ησ ≡ mσ/3H is the effective mass of the curvaton relative to the Hubble scale during inflation. Both the curvature and isocurvature perturbation (if any) originate from initial curvaton perturbation and are thus completely correlated, independent of the physics of the curvaton decay. In the absence of any additional source for the isocurvature perturbations, the curvature and any isocurvature perturbations must share the same spectral tilt, given by Eq. (49). Note that the curvature transfer coefficient r can also be related to the non-Gaussianity of the primordial perturba- tions. If r is small then the initial curvaton perturbation ζσ,in must be correspondingly large in order to produce a given amplitude of primordial curvature perturbation. This leads to a larger contribution from second-order perturbations in the curvaton density, which shows up in the non-linearity parameter [9, 29] 5 f ≈ , (50) NL 4r for r ≪ 1 [14]. (See [30] for corrections when r ∼ 1.) There is thus a possible consistency test of this curvaton model using two observables. Existing constraints on the relative amplitude of primordial isocurvature to curvature perturbations [7] constrain r> 0.9 in models where the CDM is directly produced by curvaton decay, and hence limit the allowed non-Gaussianity. However in curvaton models which produce purely adiabatic perturbations (Snuc = 0), determinations of the non-linear parameter fNL may be the only way to determine independently both the transfer parameter r (and hence the model parameter p) and the initial amplitude of the curvaton perturbation ζσ. 13 Acknowledgments The authors are grateful to David Lyth and Lorenzo Sorbo for useful comments. This work was supported by PPARC grants PPA/G/S/2000/00115, PPA/G/S/2002/00098, and PPA/V/S/2001/00544. KAM was supported by a Marie Curie Fellowship under contract number HPMF-CT-2000-00981. DW is supported by the Royal Society. [1] R. Trotta, A. Riazuelo and R. Durrer, Phys. Rev. Lett. 87, 231301 (2001) [arXiv:astro-ph/0104017]. [2] M. Tegmark et al. [SDSS Collaboration], arXiv:astro-ph/0310723. [3] C. Gordon and K. A. Malik, arXiv:astro-ph/0311102. [4] P. Crotty, J. Garcia-Bellido, J. Lesgourgues and A. Riazuelo, Phys. Rev. Lett. 91, 171301 (2003) [arXiv:astro-ph/0306286]. [5] H. V. Peiris et al., Astrophys. J. Suppl. 148, 213 (2003) [arXiv:astro-ph/0302225]. [6] L. Amendola, C. Gordon, D. Wands and M. Sasaki, Phys. Rev. Lett. 88, 211302 (2002) [arXiv:astro-ph/0107089]. [7] C. Gordon and A. Lewis, Phys. Rev. D 67, 123513 (2003) [arXiv:astro-ph/0212248]. [8] J. Valiviita and V. Muhonen, Phys. Rev. Lett. 91, 131302 (2003) [arXiv:astro-ph/0304175]. [9] E. Komatsu et al., Astrophys. J. Suppl. 148, 119 (2003) [arXiv:astro-ph/0302223]. [10] A. R. Liddle and D. H. Lyth, “Cosmological Inflation And Large-Scale Structure”, CUP (2000). [11] S. Mollerach, Phys. Rev. D 42 (1990) 313. [12] A. D. Linde and V. Mukhanov, Phys. Rev. D 56, 535 (1997) [arXiv:astro-ph/9610219]. [13] K. Enqvist and M. S. Sloth, Nucl. Phys. B 626, 395 (2002) [arXiv:hep-ph/0109214]. [14] D. H. Lyth and D. Wands, Phys. Lett. B 524, 5 (2002) [arXiv:hep-ph/0110002]. [15] T. Moroi and T. Takahashi, Phys. Lett. B 522, 215 (2001) [Erratum-ibid. B 539, 303 (2002)] [arXiv:hep-ph/0110096]; Phys. Rev. D 66, 063501 (2002) [arXiv:hep-ph/0206026]. [16] K. Dimopoulos and D. H. Lyth, arXiv:hep-ph/0209180; K. Enqvist and A. Mazumdar, Phys. Rept. 380, 99 (2003) [arXiv:hep-ph/0209244]. [17] N. Bartolo and A. R. Liddle, Phys. Rev. D 65, 121301 (2002) [arXiv:astro-ph/0203076]. D. Wands, N. Bartolo, S. Matarrese and A. Riotto, Phys. Rev. D 66, 043520 (2002) [arXiv:astro-ph/0205253]. M. Postma, Phys. Rev. D 67, 063518 (2003) [arXiv:hep-ph/0212005]. M. Giovannini, Phys. Rev. D 67, 123512 (2003) [arXiv:hep-ph/0301264]. [18] D. Wands, K. A. Malik, D. H. Lyth and A. R. Liddle, Phys. Rev. D 62, 043527 (2000) [arXiv:astro-ph/0003278]. [19] D. H. Lyth and D. Wands, arXiv:astro-ph/0306498. [20] D. H. Lyth, C. Ungarelli and D. Wands, Phys. Rev. D 67, 023503 (2003) [arXiv:astro-ph/0208055]. [21] D. H. Lyth and D. Wands, arXiv:astro-ph/0306500. [22] K. A. Malik, D. Wands and C. Ungarelli, Phys. Rev. D 67, 063516 (2003) [arXiv:astro-ph/0211602]. [23] H. Kodama and M. Sasaki, Prog. Theor. Phys. Suppl. 78 (1984) 1. [24] J. M. Bardeen, P. J. Steinhardt and M. S. Turner, Phys. Rev. D 28, 679 (1983). [25] J. M. Bardeen, DOE/ER/40423-01-C8 Lectures given at 2nd Guo Shou-jing Summer School on Particle Physics and Cosmology, Nanjing, China, Jul 1988. [26] Y. Nambu and A. Taruya, Prog. Theor. Phys. 97, 83 (1997) [arXiv:gr-qc/9609029]; F. Finelli and R. H. Brandenberger, Phys. Rev. Lett. 82, 1362 (1999) [arXiv:hep-ph/9809490]; A. R. Liddle, D. H. Lyth, K. A. Malik and D. Wands, Phys. Rev. D 61, 103509 (2000) [arXiv:hep-ph/9912473]. [27] M. Bucher, K. Moodley and N. Turok, Phys. Rev. D 62, 083508 (2000) [arXiv:astro-ph/9904231]. [28] D. Wands, N. Bartolo, S. Matarrese and A. Riotto, Phys. Rev. D 66, 043520 (2002) [arXiv:astro-ph/0205253]. [29] E. Komatsu and D. N. Spergel, Phys. Rev. D 63, 063002 (2001) [arXiv:astro-ph/0005036]. [30] N. Bartolo, S. Matarrese and A. Riotto, arXiv:astro-ph/0309692. [31] J. M. Bardeen, Phys. Rev. D 22 (1980) 1882. [32] J. Garcia-Bellido and D. Wands, Phys. Rev. D 53, 5437 (1996) [arXiv:astro-ph/9511029]. APPENDIX A: GAUGE-INVARIANT ADIABATIC AND ENTROPY PERTURBATIONS In a multi-component system it is often useful to re-write the coupled perturbation equations in terms of the instan- taneous overall curvature perturbation (often called the adiabatic perturbation) and the relative entropy perturbations between fluids. In this appendix we follow the treatment of Ref. [22] and the reader is referred to this source for more details. Throughout this paper we work in the large scale limit, i.e., neglecting gradient terms. 14 1. Evolution of curvature and entropy perturbations The gauge-invariant definition of the total curvature perturbation on uniform-density hypersurfaces is [18, 24, 31] δρ ζ ≡−ψ − H , (A1) ρ˙ where ψ and δρ are the gauge-dependent curvature and density perturbations. The evolution of the curvature perturbation is given on large scales by [18, 22, 32] H ζ˙ = − δP , (A2) ρ + P nad 2 2 ˙ where the non-adiabatic pressure perturbation is δPnad ≡ δP − cs δρ and the adiabatic sound speed is cs = P/ρ˙. Thus the total curvature perturbation is constant on large scales for purely adiabatic perturbations. In the presence of more than one fluid, the total non-adiabatic pressure perturbation, δPnad, may be split into two parts, δPnad ≡ δPintr + δPrel . (A3) The first part is due to the intrinsic entropy perturbation of each fluid δPintr = δPintr,α , (A4) α X where the intrinsic non-adiabatic pressure perturbation of the α-fluid is given by 2 δPintr,α ≡ δPα − cαδρα , (A5) 2 ˙ where cα ≡ Pα/ρ˙α is the adiabatic sound speed of that fluid. The second part of the total non-adiabatic pressure perturbation (A3) is due to the relative entropy perturbation between different fluids, 1 δP ≡− ρ˙ ρ˙ c2 − c2 S , (A6) rel 6Hρ˙ α β α β αβ α,β X where Sαβ is the relative entropy (or isocurvature) perturbation, Eq. (28), δρ δρ S = −3H α − β . (A7) αβ ρ˙ ρ˙ α β Note, that Eq. (A6) corrects a sign error in Eq. (2.31) of Ref. [22]. The evolution equation for the relative entropy perturbation Sαβ on large scales is given by [22] 3HδP − δQ − δQ 3HδP − δQ − δQ S˙ =3H intr,α intr,α rel,α − intr,β intr,β rel,β , (A8) αβ ρ˙ ρ˙ α β where the intrinsic non-adiabatic energy transfer perturbation is defined as [22] Q˙ α δQintr,α ≡ δQα − δρα , (A9) ρ˙α and the relative non-adiabatic energy transfer is defined as Qαρ˙ δρα δρ Qα δQrel,α = − = − ρ˙βSαβ . (A10) 2ρ ρ˙α ρ˙ 6Hρ Xβ For fluids, such as radiation or non-relativistic matter in our curvaton model, with definite equation of state, Pα(ρα), the intrinsic non-adiabatic pressure perturbation in Eq. (A4) is zero. Hence the total non-adiabatic pressure perturbation (A6) for our model can be written as ρ˙ δP = δP = γ (ρ ˙ S +ρ ˙ S ) . (A11) nad rel 9Hρ˙ m mγ σ σγ 15 Hence the evolution equation (A2) for ζ is Hρ˙ ζ˙ = γ (ρ ˙ S +ρ ˙ S ) . (A12) 3ρ ˙2 σ σγ m mγ The intrinsic energy transfer perturbation, following from Eq. (21), are δQintr,σ = 0 , (A13) Γ δQ = − 1 ρ˙ S , (A14) intr,γ 3H σ σγ Γ δQ = − 2 ρ˙ S , (A15) intr,m 3H σ σm while the relative energy transfer perturbations are (Γ +Γ )ρ δQ = 1 2 σ (ρ ˙ S +ρ ˙ S ) , (A16) rel,σ 6Hρ γ σγ m σm Γ ρ δQ = 1 σ (ρ ˙ S +ρ ˙ S ) , (A17) rel,γ 6Hρ σ σγ m mγ Γ ρ δQ = 2 σ (ρ ˙ S − ρ˙ S ) . (A18) rel,m 6Hρ σ σm γ mγ Hence the evolution of the relative entropy perturbations can be written as 1 ρ˙ ρ˙ ρ˙ ρ˙ S˙ = Γ ρ m S − (Γ +Γ )ρ m S − 2Γ ρ σ − Γ ρ γ S , (A19) σγ 2ρ 1 σ ρ˙ mγ 1 2 σ ρ˙ σm 1 ρ˙ 2 σ ρ˙ σγ γ σ γ σ ρ˙ ρ˙ ρ ρ ρ˙ ρ ρ˙ S˙ = Γ m − Γ γ σ S +Γ 1 − σ σ S − Γ 1 − σ σ S , (A20) mγ 1 ρ˙ 2 ρ˙ 2ρ mγ 2 2ρ ρ˙ σm 1 2ρ ρ˙ σγ γ m m γ ρ ρ˙ ρ ρ˙ ρ ρ˙ ρ ρ˙ S˙ = −(Γ +Γ ) σ γ S − Γ 1 − σ σ + (Γ +Γ ) σ m S − Γ σ γ S . (A21) σm 1 2 2ρ ρ˙ σγ 2 2ρ ρ˙ 1 2 2ρ ρ˙ σm 2 2ρ ρ˙ mγ σ m σ m Note that we can always identify a trivial solution to Eqs. (A19–A21) with all the entropy perturbations Sαβ =0 which corresponds to the adiabatic mode and leaves ζ = constant in Eq. (A12), even in a system such as this including energy transfer between different components. 2. Singular and Non–Singular Hypersurfaces From Eq. (A7) we see that the relative entropy perturbation Sαβ becomes singular wheneverρ ˙α = 0. This can occur if we have multiple fluids with energy transfer between them whenever Qα = 3H(ρα + Pα). This is because the uniform α-fluid hypersurfaces become ill-defined to first-order whenever ρα is stationary, i.e., the first-derivative vanishes. Thus a divergence in a gauge-invariant quantity such as Sαβ need not signal the breakdown of perturbation theory, but instead may indicate that a particular choice of hypersurface was not well-defined [26]. This is indeed the case in the curvaton model studied in this paper: we see that the density perturbations on uniform-curvature hypersurfaces remain finite even whenρ ˙α = 0. From the definition of Sαβ, Eq. (A7) above, we see that we can construct a non-singular quantity ∆αβ ≡ ρ˙αρ˙βSαβ = −3H (ρ ˙βδρα − ρ˙αδρβ) . (A22) Since the definition of Sαβ is gauge invariant then so is ∆αβ. The evolution equation for the total curvature perturbation on large scales, Eq. (A2) can then be written in terms of ∆αβ as 1 ζ˙ = HδP − c2 ∆ . (A23) 3H(ρ + P )2 intr,α α αβ α X Xβ The uniform total density hypersurfaces, on whichζ is defined, always remain well defined, sinceρ ˙ 6= 0 after inflation.